[00:52] <fretegi_away> thanks guys
[00:52] <fretegi_away> bbiab later tonight to tackle again
[00:52] <fretegi_away> good thing is, one drive of array is working and its all backed up as of an hour ago ha
[11:58] <geodb27> People : hi ! I'm trying to follow this document : https://ubuntu.com/server/docs/install/autoinstall-quickstart to create a customized auto-install cd of 20.04 lts, however, I'm stuck with the iso creation. The example shown deals with a vm hosted on a computer and launched via kvm. This is not what I have to do : I must create vms with vcenter/vmware. So far as I know, I can only mount an iso and instruct to boot on it. So, how can I get
[11:58] <geodb27> an automated install on this infrastructure ?
[12:00] <rbasak> Do you have a reason for not just using a cloud image?
[12:02] <Ussat> you need to use the esxi comnmand line version of the kvm --boot option
[12:03] <Ussat> and...there are better ways
[12:09] <geodb27> Ussat: If there is a better and easier way to achieve that, then I'll take it. I have around 20 machines to create, each with the same partionning schema. Basically, the only changes between these 20 vms are their name and IP address.
[12:10] <geodb27> rbasak: I've setup a machine manually and have all the yaml descriptions I need in /var/log/installer/curtin-install-cfg.yaml so, I guess that I can use part of it as a base, indeed.
[12:13] <Ussat> On esxi build one system exactly how you want it, then make clones
[12:13] <Ussat> Thats exactly what templates are for
[12:17] <geodb27> This could be a solution, indeed. But that's not the way we process. I've written that these 20 machines would be quite the same, this is not true as for the cpu/ram/disks.
[12:19] <Ussat> so ? once the systems are created use esxi to modify that
[12:20] <Ussat> Basically, either you want clones or not, esxi does not have the ability to use cli tools lkike you want
[12:20] <Ussat> or, use KVM
[12:32] <geodb27> Ussat: I know it, but for historical reasons, we maintain our vm this way : one kind-of kickstart (redhat/ubuntu) to primo install our vms. No use of esxi templates or the like. So, I must conform to what I'm expected to do.
[12:32] <Ussat> ..
[12:32] <Ussat> Um.....ok.
[12:32] <Ussat> Not a good reason, but gluck then
[12:34] <geodb27> That said, is there a way to give the preseeded file (cloud-init syntax) as one did with 18.04 version of lts ? Some kernel parameter to feed ?
[12:46] <rbasak> Then use kvm to prepare a cloud image - no installer needed - and then hand that image over to vmware?
[13:11] <floogy> Hi, I can't get wifi with netplan working on 20.04 raspberry Pi 3B+
[13:28] <floogy> Uses netplan /etc/wpa_supplicant/wpa_supplicant.conf to create /run/netplan/wpa-wlan0.conf ?
[13:30] <floogy> I got a yaml lint validated /etc/netplan/50-cloud-init.yaml with awifi section. I ran sudo netplan try, sudo systemctl daemon-reload and sudo netplan apply.
[13:36] <floogy> reason=CONN_FAILED
[13:37] <floogy> wlan0: CTRL-EVENT-ASSOC-REJECT bssid=38:10:d5:73:45:3a status_code=16
[13:38] <floogy> IPv6: ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
[13:39] <floogy> I don't know at all why it gets rejected.
[13:42] <floogy> https://pastebin.pl/view/78064dfa
[13:49] <floogy> Formerly I used raspbian (debian stretch) without issues with wpa_supplicant. the microSD died, and I saved it with gddrescue. fsck found merely no errors but a journal it could restore. Also vfat wasn't affected by the thousands of badblocks and -adareas. Now I mounted the image and looked into the wpa_supplicant.conf and found nothing special.
[13:51] <floogy> Using these examples: https://netplan.io/examples/#connecting-to-an-open-wireless-network
[13:51] <floogy> http://www.yamllint.com/
[14:00] <floogy> Do I have to use hostapd? https://askubuntu.com/questions/1222278/detailed-how-to-or-example-needed-to-setup-an-access-point-using-netplan
[14:03] <floogy> But I want to use the raspi as a member in my LAN that joined the AVM Fritz!Box Router as accesspoint ...
[14:47] <geodb27> Hi again. I'm trying to understand this document : https://ubuntu.com/server/docs/install/autoinstall-quickstart In particular the second Chapter "Run the install!" The kvm command given instructs the new vm to be build against a 10G disk (image.img), to use a second disk (seed.iso) and to boot (by which magic, this is not explained) on the mounted iso image as a cdrom. How is this method configured ? How does the install process know where
[14:47] <geodb27>  to mount/load the seed.iso ?
[14:55] <rbasak> geodb27: the installer is intended for bare metal, so you would boot bootable media just the same as you do normally.
[14:55] <rbasak> You said that you're using vmware. In that case "how to boot an ISO" is a vmware question and not specific to Ubuntu.
[15:06] <geodb27> rbasak: thanks for your answer, but I fear that you misunderstood my question. So I will try to elaborate : I just wanted to know how the installer is instructed to search the /dev/sdb for cloud-init files. I know how to boot an iso in vmware and that was not my last question. Sorry for the misunderstanding.
[15:15] <rbasak> geodb27: ah. I think that's maybe https://cloudinit.readthedocs.io/en/latest/topics/datasources/configdrive.html
[15:15] <rbasak> cloud-init scans all drives for those criteria AIUI
[15:16] <rbasak> Sorry that's probably the wrong link
[15:16] <rbasak> I meant https://cloudinit.readthedocs.io/en/latest/topics/datasources/nocloud.html
[15:18] <geodb27> Thank you so much rbasak, I think that with that, I'll be abble to do what I want !
[18:45] <Thumpxr> Anyone got a clue, how i can "cut" my own domain from a log file which fail2ban parses?
[18:51] <Thumpxr> got it ... it's not cutting/removing but rather writing the regex so that it checks the hostname later. >> failregex = mydomain.tld.* <HOST>.*POST.*(wp-login\.php|xmlrpc\.php).* 200
[20:09] <tafa2> How do you guys do file level backups on *nix production servers? I'm trying to find a pull based solution/method with deduplication. Any recommendations?
[20:35] <tomreyn> borg and restic, and (very different) bacula, are the backup systems i see most recomended. i'm not sure whether any of them support pull (bacula probably does).
[20:42] <tafa2> tomreyn bacula does, borg and restic don't natively
[20:42] <tomreyn> so... you already know?
[21:27] <floogy> Is it a good idea to mount /var per nfs to reduce write access on microSD cards on the raspberry Pi? Maybe with a script, that switches to tmpfs if the netdrive isn't available?